Capturing cell cycle activity in the live mouse cornea.

(A) Genetic alleles of in vivo cell cycle reporter (Cyclin B1-GFP) used to sort stem and progenitor cells in the corneal epithelium. Cell cycle diagram displays how brightness of the GFP reporter changes throughout the cell cycle. (B) Individual planes from a series of optical sections of the mouse cornea taken at the indicated depths. (Also see Movie S1.) (C) Representative snapshots from time-lapse imaging of the cornea in a live CB1-GFP mouse. Only actively cycling cells are labelled with GFP (Also see Movie S2). (D) Reconstructed side view (XZ) of limbus and cornea from a CB1-GFP mouse based on 2-photon intravital imaging. (E) Panel shows representative snapshots from a time course live imaging experiment. The limbus and peripheral cornea were imaged at high magnification at the indicated times after treatment with TPA. Dotted line indicates the limbal-corneal border. (F-H) Optical sections of a CB1-GFP mouse eye imaged at the indicated epithelial compartments. The globally expressed nuclear tomato reporter allowed for visualization and quantification of total cell numbers. Graphs show quantification of the number of CB1-GFP+ cells (top panel) and percentage of CB1-GFP+ cells in each epithelial compartment (n = 6 [conjunctiva] or 8 [limbus + cornea] sampled images per epithelial compartment from 2 mice, one-way ANOVA). *p < 0.05, **p <0.01, ***p < 0.001, ****p < 0.0001. (I) Representative frames from time-resolved live imaging series, showing CycB1-GFP+ cells located in the basal layer of the limbus and central cornea. Arrowheads denote the axis of the daughter cell separation during mitosis, which are quantified and shown ass radial graphs (right panel) p = 0.86 (n = 35 cells [limbus] and 35 cells [limbus] sampled from 6 time-lapse datasets.

Single cell transcriptomic analysis of corneal stem and progenitor cells.

(A) Experimental design for isolating and profiling actively cycling cells in the mouse ocular epithelium. (B) UMAP plot presentation of GFP+ cells sorted from corneas of CB1-GFP mice. Unbiased clustering of 5545 cells (2 pooled samples, 30 and 22 corneas total) revealed 8 distinct groups: Conjunctiva (860 cells), Limbus (LSCs; 823 cells), TAC (goblet niche; 150 cells), TACs (adaptive; 702 cells), TACs (peripheral; 1443 cells), TACs (central; 875), TACs (terminal; 668 cells), Immune (24 cells). (C) Stacked violin plots of differentially expressed markers used to define cluster identity. The y-axes represent expression level. (D) Inference trajectory analysis. Dotted lines show inferred pseudotime trajectories of corneal progenitors. (E) Dot plot of top differentially expressed transcription factors. (F) Umap plot for Sox9. (G) Volcano plot comparing differentially expressed genes between Sox9-high (>1.3 log) or low (<1 log) expressing cells.

Sox9 marks long-lived limbal stem cells that maintain the cornea.

(A) Genetic alleles for in vivo imaging. (B) Global and high-magnification views of the eye imaged at the indicated epithelial compartments show GFP expression predominantly at the limbus (n = 3 animals). Scale bars represent 500µm (global view) and 100µm (high magnification). (C) In vivo lineage tracing of Sox9CreER;tdTom cells by longitudinal live imaging. Representative tracing time series generated by reimaging the same eye at the indicated time points after tamoxifen administration (n = 5 mice). Lower panels show representative magnified views of the same compartment. Dotted lines indicate the margins of the limbus. Scale bars represent 500µm (panel B and 5x panels) and 250µm (panel C and10x panels). Abbreviations: DPI, Days Post Induction.

Sox9 cKO corneas develop squamous metaplasia.

(A) Experimental timeline. Adult mice (p63CreER; Sox9 fl/fl cKO and control littermates) were treated with Tamoxifen daily for 3 days and analyzed 2 months post tamoxifen induction (MPI). (B-I) Gross morphology and representative histological sections of the central cornea (D, H) and limbus (E, I) of a control and Sox9-cKO cornea at 2MPI. (F-H) Quantifications of tissue thickness for the epithelium, stroma, and total corneal thickness in each epithelial compartment (control corneas: n = 13 animals [25 eyes]; Sox9-cKO corneas: n = 11 animals [21 eyes], Welch’s t test). (I-L) Representative histological sections depicting phenotype severity. Phenotype severity was scored from 1-4 as follows: 1) No gross morphological/molecular abnormalities (appear like controls), n = 7/21 corneas (6/11 animals); 2) Mild thickening of the epithelium and/or stroma, n = 7/21 corneas (6/11 animals); 3) Moderate stromal thickening with mild to moderate epithelial thickening, n = 2/21 corneas (1/11 animals); 4) Severe thickening of stroma, evidence of keratinization (loricrin expression), n = 5/21 corneas (4/11 animals). (M) Bar graph and table depicting distribution of corneas across phenotype severity categories and the proportion of corneas that developed opacities, respectively. Scale bars are 200 µm.

Cell migration and improper differentiation drive cellular crowding at the central cornea.

(A) Allelic system and experimental strategy used to generate single, Differentiation-OFF clones in the basal layer of the corneal epithelium. Tamoxifen administration in low doses enables Cre-mediated excision of loxP-flanked sequences in a reporter allele and a tetracycline transactivator allele (tTA) in Keratin14-expressing cells. Membrane-localized GFP (mGFP) becomes expressed after excision of membrane-localized tdTomato (mTom) in single clones while excision of a STOP cassette in the ROSA26 locus allows tTA expression. tTA expression in single clones enables expression of two TetO alleles: a dominant negative MAML allele to suppress Notch signaling (Differentiation-OFF) and a reporter allele (TetO-H2BGFP) to label the single “Differentiation-OFF” clones with nuclear GFP. Control animals have only the one TetO-H2BGFP reporter allele. Labelled cells were followed over time by re-imaging the same areas of the cornea. (B) Labelled clones in control animals (n = 5) were tracked differentiating into the suprabasal and superficial layers of the cornea. Differentiation-OFF (n = 6 animals) clones were observed primarily in the basal layer with very few cells observed differentiating into the suprabasal and superficial layers. (C) Left panels show representative frames from a live imaging timecourse, showing the dynamics of control and Differentiation-OFF basal clones. Quantitative clonal analysis is shown in the graphs on the right. Two-way ANOVA, P<0.0001, n = 10 clonal fields tracked in 4 mice. (D) Representative frames from a live imaging timecourse of the central cornea from a Differentiation-OFF mouse. Over time, clones could be observed clonally expanding and migrating radially towards the central cornea. (E) Representative examples of corneas from Differentiation-OFF mice that developed squamous metaplasia in the central cornea.

Mechanistic Model for the role of Sox9 in the regulation of stem cell fate in the ocular epithelium.

Single cell sequencing reveals CycB1-GFP sorted cells from the corneal epithelium have a mitotic, basal cell signature. (A) Representative images of FACS gating strategy used to sort for GFP+ cells from Cyclin B1-GFP animals. (B) Umap plots of integrated single cell datasets comparing this study with a published study of single cell transcriptomic analysis of bulk corneal epithelial cells (Altshuler et al. 2021). (C) Violin plots of proliferation (Mki67) and differentiation (Dsg1a) markers. (D) Violin plots of basal, suprabasal, cell cycle and apoptotic markers of mouse corneal epithelial cells profiled in the present study.

(A) Dot plot of top differentially expressed genes between the identified cell clusters. (B) Umap plots of differentially expressed genes in each of the identified cell clusters.

Sox9 Expression in the mouse eye. (A/A’) Immunofluorescence images of Sox9 in the limbus and central cornea (C/C’) of control corneas (n = 25 corneas [13 animals]). (B/B’) Immunofluorescence images of Sox9 in the limbus and central cornea (D/D’) in Sox9-cKO corneas (n = 21 corneas [11 animals]). (E/E’) Immunofluorescence images of control and Sox9-cKO (F/F’) eyes stained for Sox9 in the retina. Muller glial cells in the retina are non-epithelial cells that also express Sox9 but are not targeted for deletion by Cre recombinase. Asterisks indicate non-specific background staining. Scale bars are 100 µm.

Keratin expression is abnormal in Sox9 cKO corneas. (A/A’) Immunofluorescence images of K14 in the limbus and central cornea (C/C’) of control corneas (n = 25 corneas [13 animals]). (B/B’) Immunofluorescence images of K14 in the limbus and central cornea (D/D’) of Sox9 cKO corneas (n = 21 corneas [11 mice]). (E/E’) Immunofluorescence images of K12 in the limbus and central cornea (G/G’) of control corneas (n = 25 corneas [13 animals]). (F/F’) Immunofluorescence images of K12 in the limbus and central cornea (H/H’) of Sox9-cKO corneas (n = 21 corneas [11 animals]). Keratin 14 (K14), Keratin 12 (K12); Scale bars are 100 µm.

Sox9 cKO corneas express the cornification marker, loricrin. (A/A’) Immunofluorescence images of Loricrin in the limbus and central cornea (C/C’) of control corneas (n = 25 corneas [13 animals]). (B/B’) Immunofluorescence images of Loricrin in the limbus and central cornea (D/D’) in Sox9-cKO corneas (n = 21 corneas [11 animals]). Loricrin (LOR); Asterisks indicate non-specific background staining. Scale bars are 100 µm.

Sox9 loss leads to unchecked proliferation throughout the corneal epithelium. (A/A’) Immunofluorescence images of KI67 in the limbus and central cornea (C/C’) of control corneas (n = 8 corneas [7 mice]). (B/B’) Immunofluorescence images of KI67 in the limbus and central cornea (D/D’) in Sox9-cKO corneas (n = 7 corneas [4 animals]).